i915 issue

[decodering22]

After boot kernel panic and can't install DSM due to i915 issue. Is the i915 built into this latest build?

[PeterSuh-Q3]

SA6400 ?

[decodering22]

Hi Peter yes SA6400, been pulling my hair trying to get xpenology installed and no luck

[PeterSuh-Q3]

BareMetal or VM like Proxmox?
And what is the CPU model?

[decodering22]

Baremetal intel i5-12500 on MSI MGP 760I motherboard. If there is anything you want me to try let me know I'm in front of it now, been battling this for a few hours

[PeterSuh-Q3]

There is no space in the message box of the Henology forum. I wish we would only talk on GitHub.

Are all ARC/RR/mshell in the same situation?

https://xpenology.com/forum/topic/60130-redpill-tinycore-loader-installation-guide-for-dsm-71-baremetal/?do=findComment&comment=456343

[decodering22]

I only tried ARC and RR. Initialy i tried RR and it just kernel panic so assumption is i915 is the issue but I'm not sure. I tried your RR build and after the QR code pops up then screen goes blank. I assumed based on that thread the latest i915 was loaded on i915 so it should work. Not sure what it could be b/c there is no DSM log since I can't even install DSM.

do you want to email or even call me?

[PeterSuh-Q3]

https://xpenology.com/forum/topic/60130-redpill-tinycore-loader-installation-guide-for-dsm-71-baremetal/?do=findComment&comment=456343

Looking at this log, it appears that mshell has already been tried.

Are you sure that i915 is causing the panic by checking the logs?
It's hard to be sure until you check dmesg yourself.
rr and mshell provide functions to retrieve and analyze dmesg from error situations.

The fact that the screen disappears on SA6400 may mean that i915.ko is working well.
The bug where the screen disappears after i915.ko is loaded properly is also confirmed in mshell.

It seems to me that the Realtek 2.5G network is not operating stably.
Are there any additional NICs that can be installed? It's better if it's from Intel.

[decodering22]

I have a 10Gbe intel card i plan to use you think if i disable the onboard Realtek that will fix the issue? I can disable it and test it for sure that is not a problem i will disable it on the UEFI bios.

I can't check the log b/c as soon as the IP loads it causes the kernel panic. The only reason i went down the i915 rabbit hole is b/c it was mentioned that was the issue. Let me pop in the intel X520-DAC1 and disable the Realtek 2.5G to see if that works or not.

Where can i pull the log to be 100% sure?

[decodering22]

I also have a 1GBe intel card i will pop in for the test as well

[decodering22]

Same thing, i disabled the Realtek 2.5Gbe NIC in the BIOS itself, it detected the new pcie 1Gbe Intel VRRH1-HIGH Nic without issues. and the result is the same. Screen goes black and after about 15 seconds I can ping it again but can't be found in finds.synology.com or the IP with port to connect to it.

Can you tell me where to pull the log from and when. Really appreciate your help

[decodering22]

If there is some settings you want me to try and test let me know. Not sure how to pull the log b/c it kernel panics right away

[PeterSuh-Q3]

In the case of mshell, the last log file is saved when the Friend kernel is loaded.
It exists in the directory below.
/mnt/tcrp-p1/logs/jr
Friend, stop with ctrl+c while booting.
Please examine the files under this directory.
You should also look at dmesg, but the contents of linuxrc.syno.log are also important.
Can you provide me with these 2 files for download?

[decodering22]

Yes let me load RR on a flash drive right now and get these to you now. Give me a min I have to go through the entire process probably a few mins to get it to the point where it does this. Do you want me to email it to you directly or do you want to continue here on github?

[PeterSuh-Q3]

I gave an example of MSHELL. I don't know exactly where and how RR's logs are stored.
GitHub is also good enough.

[decodering22]

I have your tinycore-redpill that is what I meant when i said RR not sure if your reference to RR is something else but i assume the redpill is the MSHELL which is what I loaded prior and now

[decodering22]

you are correct i can't get access to this repo https://github.com/wjz304/arpl-i18n/releases

If you have it i can try it

[PeterSuh-Q3]

Use this
https://github.com/syno-community/rr/releases

[decodering22]

Thanks Peter i will try this as well. I ordered a new motherboard as well and it should be here today. I took the other system apart already so when the new one arrives I will try this on it.

[PeterSuh-Q3]

These are successful cases using Intel 12th generation and SA6400, including N100, at the server forum in Korea where I work. It would have been better if you had let me know in advance.
good luck.

https://svrforum.com/?act=&vid=&mid=case&category=&extra_vars5=12&extra_vars6=&extra_vars7=&extra_vars8=&extra_vars9=&extra_vars10=

[decodering22]

Thanks Peter I'll translate the site and see. It has to be some compatibility with the Motherborad i was using. I'm still waiting for the new one to arrive, it does not have any Realtek NIC, only Intel, but still not sure if that was even the issue since we couldn't capture logs before it crashes.

I'll let you know as soon as I have the new system to see if it works.

[decodering22]

Hi Peter, so I had some luck with this new motherboard and it is working but an odd issue is if i remove the USB drive it just goes straight to the BIOS. It seems like i have to have the USB drive plugged in all the time so it forces DSM to boot. Is this normal? Is there a way to make the SATA drive bootable so no USB is needed?

[decodering22]

Peter just so i understand the loader needs a USB drive to boot, than it installs the DSM on the SATA SSD correct. Is it possible to partition part of the SATA SSD so it is used to boot the DSM and act like the loader?

[PeterSuh-Q3]

It's impossible. You need to understand the mechanism by which Synology disks' boot devices and data disks work. All data disks have the same system partition at the front of the disk, and DSM is installed equally there. DSM itself does not support booting. Genuine devices required for Synology booting use durable Sata Dom. Redpill loaders such as TCRP and ARPL create boot loaders with the same structure as this sata dom. Small capacity USB, MSata, SSD, etc. must be made independent with a separate bootloader. M.2 Sata SSD is also available, but Nvme is not available.

[decodering22]

If i understand the USB or SATA DOM is needed no matter what. The SATA disk and M.2 based on what you stated both have the same DSM system partition. So if I add a second M.2 drive its not just a Volume drive only, but there is a small system partition as well correct? I'm going to just use a small low profile and fast USB and call it a day. I was hoping to avoid it but it doesn't sound like it is possible.

It is only used during boot only correct, so minimal Read/Write on the USB? I don't reboot often so I suspect it should be fine?

[PeterSuh-Q3]

There are two types of M.2, so they must be clearly distinguished.
These are M.2 Sata SSD and M.2 NVMe SSD.
If you add an M.2 Sata SSD, it operates the same as a regular SATA (SSD/HDD) disk.
Any Sata based disk can be used as a bootloader.
However, M.2 NVMe SSD can only exist as a cache, but with the help of a special addon, it can be turned into a data disk volume. It cannot be used as a bootloader.

USB sticks are the most vulnerable devices in terms of durability.
It stays on the PC during booting and does not participate in reading/writing, but
Must be exposed to continuous heat.
This is a factor that reduces the durability of the USB stick.

[decodering22]

Hi Peter, great explanataion, I know the difference between the two types of M.2 drives and how they are utilized in xpenology pradigm with m.2 as cache but with the drive DB update being able to be used as data disk volume.

Tracking that M.2 NVMe PCIe can't be used for bootloader and right now it isn't fully supported without a SATA drive.

What I was concerned with was the USB drive being suseptiple to failure and if it was possible to partition the SATA SSD drive for a boot partition and the other partition being used for the DSM System partition. It sounds like that is not possible at the moment and the use of a USB or separate SSD(SATA) or SATA DOM is needed for the boot loader alone.

I was under the impression with a previous version (Jun's) release in DSM 6.x days he had figured out a way to use the SSD (SATA) as both a bootloader and DSM system partition too. I just assumed that was also possible now with DSM 7.2 and beyond.

In your opinion is it worth it to transition to a SATA DOM for long term reliability over USB?

[PeterSuh-Q3]

I've seen users using SATA Doms instead of actual USB disks. I haven't directly encountered verified data on durability, but isn't the fact that Synology is using it enough verification?

[decodering22]

Yea if Synology is using a SATA DOM than it must be reliable enough. I don't mind getting a small cheap SATA drive. I'm just concerned the USB disk failing overtime like you mentioned due to heat. The only options for SATA DOM are supermicro i have found and they don't have the greatest review. I have plenty of SSD lying around and there is plenty of open SATA slots. For now I ordered a Sandisk Fit and will use that until it fails than transfer to another option. Thanks for your help

[PeterSuh-Q3]

/var/log/linuxrc.syno.log

The file shows most of what REDPILL does.
However, the clean log in SA6400 is due to the characteristics of the EUDEV method.
In the EUDEV method, most important parts of the log are complex enough to be judged by dmesg.
As you can see, dmesg contains quite a bit of content.

Whether differences in BIOS versions really cause kernel panics or other errors can be determined only through logs.

To facilitate collection of this log, the NIC must be operating at a minimum level.

The most serious problem is when even the NIC does not work, and it becomes difficult to determine the cause.

If only the NIC is working, use a web browser
IP address: 7681 via port connection
You can check the causes of the problem using /var/log/linuxrc.syno.log or dmesg.

An alternative for when the NIC is not working is console monitoring through the Serial COM1 port.
Since the Serial COM1 port is no longer supported by the latest MOBO, it is currently viewed as an unusable solution.

I am using the dbgutils addon that is used with rr/arc/mshell to easily check for these problems.

This is a function that stores various logs, including the two files mentioned above, which contain the situation up to a possible kernel panic during mshell booting, in a specific location and allows them to be checked separately.
It is not yet known whether this function works reliably.

The best thing is console monitoring through the Serial COM1 port.
For this purpose, I deliberately purchased an interface card with two physical serial ports.
However, the ADDRESS mapping that allows COM1 port to operate is not possible, so I have not yet been able to get it to work properly.

I'm still looking for an interface card that physically allows ADDRESS adjustment with DIP switches.

Efforts to monitor logs are still ongoing, but a stable solution has not yet been found.

[decodering22]

Hi Peter, even though I can ping and receive replies i cna't connect via IP:7681, no connection of any kind is working after the bootloader finishes loading and screen goes blank.

I don't even know what could possibly be in the BIOS causing this. It really makes no sense but given the chipset is the same and same hardware between the boards it must be a setting in the BIOS likely.

[PeterSuh-Q3]

And, i226-v is still a chipset with many issues.
There have been several igc module patches for this in the past, but
There are no more patches in progress recently.
If you do a Google search on i226-v and EEE (Energy Efficient Ethernet) issues, you will get an idea of the situation.

So I also bought a NIC for this chipset.
Although this is the case with mshell, it is case by case.
This NIC works in one MOBO but not in another MOBO.
I don't know why.
So, I would still recommend avoiding the i226-v chipset if possible.
However, the latest MOBO uses a lot of i226-v.

[decodering22]

That is interesting, but your deducing the problem is i226-v, if you remember initially i was chasing i915. Not saying the NIC can't be the problem but the MB which works has the same exact i226-v NIC. The funny thing is it also has a second 1Gbe NIC and that doesn't work, what works is the i226-v 2.5G port.

I guess the mystery of why it works in one MB vs another will remain and could be the culprit instead of something in the BIOS itself. Chris thinks the problem is the BIOS, he also thought it was i915 initially too.

I just wish i could nail down the problem. I'm sending the new MSI board back and sticking with the ASrock for now since it works.